Long failover time

[alinik]

Hi,

I was checking the failover times in vmware HA in vSphere 5, so I tried rebooting the ESXi host to see how much time it would take to start the VM on another host. It took 60 seconds, which the ideal time should be around 30 seconds I suppose.

What can be wrong? and what can I do to decrease this time?

P.S. In the cluster settings for VM monitoring, I set "1 second" for Failure interval. And the VM's image is on a shared NFS.

Thanks,

Ali

[sparrowangelste]

I havent used  this but it might assit:

http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=100051...

[chriswahl]

alinik wrote:

P.S. In the cluster settings for VM monitoring, I set "1 second" for Failure interval. And the VM's image is on a shared NFS.

The VM Monitoring section is for the HA agent on the host to monitor failures on the VMs. This is typically for situations where a VM has a blue screen or if the service fails / crashes.

In your case, since you rebooted the entire host, it's not applicable.

If the host your rebooted was the HA Master, it can take up to 35 seconds for any restarts to occur.

[alinik]

Thanks. But it is not applicable in vShpere 5 anymore.

[alinik]

I know, but in my configuration it takes about 60 seconds and I need to decrease it.

[chriswahl]

Did you hard reset the box or do a graceful reboot? It may also be worth checking the host logs to see the how the FDM handled the notifications from the poweron request.

It would probably be easiest to begin with troubleshooting by reseting a Slave node so you don't have to worry about an election process.